[aspect-devel] cylindrical rigid body rotation

[Wolfgang Bangerth]

On 02/28/2014 03:36 PM, Ian Rose wrote:
> It would be fairly easy to remove the net angular momentum as Wolfgang
> described, and I may end up doing that, though what I have at the moment takes
> a different strategy.  Instead I explicitly construct rotation vectors and
> project them out of the solution, i.e.
>
> v' = v - (v \cdot ( \omega \times r)) (\omega \times r)
>
> I don't know whether this or the angular momentum approach is preferable (or
> if there is some fatal flaw to my approach),  maybe somebody has some insight
> into that.

I don't know. If the density is constant, the two are equivalent. Either is 
fine to remove the null space. My inclination would be to remove the physical 
angular momentum. Do any of you geoscientists have preferences? Cedric?


I need to write some glue code to make sure we only call this piece of code 
when it is necessary, i.e., if the geometry and the boundary conditions allow 
for a null space.

@Ian: Can you also post your interpolate_onto_velocity_system() function? 
Where are the places where you call the functions that compute and remove the 
null space?

Best
  W.


-- 
------------------------------------------------------------------------
Wolfgang Bangerth               email:            bangerth at math.tamu.edu
                                 www: http://www.math.tamu.edu/~bangerth/